DescriptionThis dissertation comprises three essays. The first two apply textual analysis methodologies in the accounting and auditing domain and the last essay investigates materiality as a key audit component.
The first essay develops a novel measurement, named management self-efficacy, by analyzing earnings conference call transcripts. This measurement captures managers’ level of judgment about the ability of their company to achieve its goals. This essay provides empirical evidence of the negative association between the level of management self-efficacy and the company’s future financial performance. This result supports the overconfidence and obfuscation arguments. This study also introduces a natural language processing methodology that uses artificial neural networks for dictionary building.
The second essay documents the usefulness of the textual analysis of earnings conference calls in identifying accounting misreporting. This essay demonstrates that differences in tone between CEOs and CFOs can predict misstatements in financial statements. When accounting misreporting exists, the negative tone of CFOs during Q&A session increases significantly more than that of the CEOs. This essay documents the empirical evidence regarding narrative tone sensitivities based on a person’s job title and related personal characteristics.
The third essay concerns the materiality level, an important aspect of auditing. An external auditor sets a materiality level at an early stage of the auditing process and the subsequent audit procedures and results are directly affected by this level. However, due to the scarcity of archival data, few studies have considered the determinants of the materiality level. This essay provides empirical evidence that both new clients and long-tenured clients have a higher materiality level than other clients. Furthermore, the materiality level is negatively associated with the modified audit opinion when it is a new client.